datvuthanh / HybridNets

HybridNets: End-to-End Perception Network
MIT License
594 stars 121 forks source link

Reproduce training result #85

Open PurvangL opened 1 year ago

PurvangL commented 1 year ago

HI @datvuthanh @xoiga123 , Could you please share parameters that you use and also training strategy, to reproduce the result? I am trying to reproduce result with suggested dataset from scratch. Pricision value seems to be really bad after 5th epoch.

image

command

python train_ddp.py -p hno -c 3 -n 4 -b 32 --lr 0.001 --optim adamw --num_epochs 200 --num_gpus 8 --save_interval 1500 --freeze_seg True

Thank you